|Publication number||US5416849 A|
|Application number||US 07/964,225|
|Publication date||May 16, 1995|
|Filing date||Oct 21, 1992|
|Priority date||Oct 21, 1992|
|Publication number||07964225, 964225, US 5416849 A, US 5416849A, US-A-5416849, US5416849 A, US5416849A|
|Original Assignee||International Business Machines Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (13), Referenced by (106), Classifications (10), Legal Events (5)|
|External Links: USPTO, USPTO Assignment, Espacenet|
1. Technical Field
The invention disclosed broadly relates to data processing systems and methods and more particularly relates to techniques for the extraction of field images from scanned document images.
2. Related Patent Applications
This patent application is related to the copending U.S. patent application Ser. No. 07/870,129, filed Apr. 15, 1992, entitled "Data Processing System and Method for Sequentially Repairing Character Recognition Errors for Scanned Images of Document Forms," by T. S. Betts, V. M. Carras, L. B. Knecht, T. L. Paulson, and G. R. Anderson, the application being assigned to the IBM Corporation and incorporated herein by reference, now U.S. Pat. No. 5,251,273.
This patent application is also related to the copending U.S. patent application, Ser. No. 07/870,507, filed Apr. 17, 1992, entitled "Data Processing System and Method for Selecting Customized Character Recognition Processes and Coded Data Repair Processes for Scanned Images of Document Forms," by T. S. Betts, V. M. Carras and L. B. Knecht, the application being assigned to the IBM Corporation and incorporated herein by reference.
This patent application is also related to the copending U.S. patent application Ser. No. 07/305,828, filed Feb. 2, 1989, entitled "A Computer Implemented Method for Automatic Extraction of Data From Printed Forms," by R. G. Casey and D. R. Ferguson, the application being assigned to the IBM Corporation and incorporated herein by reference, now U.S. Pat. No. 5,140,650.
3. Background Art
The above referenced copending patent applications by T. S. Betts, et al. describe the system context within which the invention disclosed herein finds application. The system disclosed by T. S. Betts, et al. defines document forms and then reads filled-in copies of those document forms which are scanned in to a digital imaging system. Each document form which is defined, includes several fields within which handwritten or typed information is to be entered. The T. S. Betts, et al. system examines the digital image of a scanned-in form to identify the form, and then locate the respective fields from which images are extracted for character recognition operations.
The process of field extraction for digital images of document forms is made difficult by the presence of preprinted background information in the form of text, boxes, and other visual prompts which have been included on the form to assist the person filling in the form. This preprinted background information must be identified in the scanned-in copy of the form and deleted from the form. This problem becomes acute when it is realized that the scanned-in form will be offset and skewed by virtue of the mechanical imprecision of the scanning device employed. Additional problems of an even more severe nature are encountered where the person filling in the form misregisters the handwritten or typed characters. Characters which overlap or go beyond the boundary of the fields will typically will not be completely extracted in prior art field extraction processes. In addition, where artifacts such as marks, creases, staples or other unintended marks which appear on the scanned-in image of the form, create still more severe problems in discriminating between the intended characters and the unintended artifact marks on the form.
It is therefore an object of the invention to automatically perform field extraction of images from document forms, in an improved manner.
It is another object of the invention to automatically perform field extraction of images from document forms whose fields have have been filled out with ambiguous marks.
The invention is a data processing method, system and computer program for extracting fields for digital images of document forms.
The system and method for field extraction, in accordance with the invention, creates a form definition for each form, which includes a form definition data set. In the form creation process, the user will mark each field which he intends to have subjected to field extraction, on a master form which he is creating. The inventive process then defines a field viewpoint structure for each field. The field viewpoint structure includes viewpoint primitive shapes such as line ends, box ends, crossed lines, V-ends, T-ends and blobs. The inventive process breaks the area within and surrounding the marked field into regions, an inner region enclosed by the marked field and eight outer regions which surround the marked field. The process then identifies the types and locations of viewpoint primitives in each region. The process then links those primitives which are appropriate, into a view relation. The field viewpoint structure is the set of all viewpoint primitives and any view relation for a field. The process then stores the coordinates of the field mark and the field viewpoint structure for each field. The collection of coordinates and field viewpoint structures for all of the fields on a form constitutes the form definition data set for a particular form. This data set is stored in the system until a particular filled-in copy of that form is received by the system for character recognition.
The second part of the invention is the process of field extraction for the fields on a scanned-in copy of the form which has been filled in. Field extraction must be preceded by a forms recognition process which identifies the form ID and which determines the skew and the offset of the scanned-in form. The field extraction process, in accordance with the invention, then begins by using the form ID to access the form definition data set. The process then examines every field on the scanned-in form. The process adjusts the coordinates where each field is expected, by using the offset values for this particular scanned-in form. The process then compares the stored viewpoint structure for a field in the data set, with the viewpoint structure of the field for the scanned-in form. For each inner and surrounding region of a field, the process determines if there are any viewpoint primitives in the data set. These are then compared with any viewpoint primitives in the corresponding region on the scanned-in form. After all regions are examined for a field on the scanned-in form, all matching viewpoint primitives are removed from the scanned-in image. Additional field analysis steps can also be performed. For example, partial removal of strange viewpoints, removal of long lines, and removal of jagged lines. In addition, overlapped pixels are retained and isolated pixels are retained if they are close enough to the field and when the skew of the scanned-in form is considered. These operations are performed for all of the fields on a scanned-in form. Then, the extracted field images are output to a character recognition process. Some character recognition processes also require the coordinates of the fields, and these can also be output from the inventive process.
The resulting field extraction of images from document forms is substantially improved in its accuracy by use of the inventive process.
These and other objects, features and advantages will be more fully appreciate with reference to the accompanying figures.
FIG. 1 which is composed of FIGS. 1B and 1C, is an overall architectural diagram of the invention.
FIG. 1A illustrates the machine generated data structure (MGDS) 50A from the forms recognition and field extraction processor
FIG. 2A illustrates the master form 10F.
FIG. 2B illustrates the master MGDS 50M corresponding to the master form 10F of FIG. 2A.
FIG. 2C illustrates the scanned form image 10' of a document form.
FIG. 2D illustrates the extracted field images 10" from the scanned form image 10' of FIG. 2C.
FIG. 3 illustrates the sequence of processing steps for the field 16' in the document form image 10' shown in FIG. 2C.
FIG. 4 illustrate the structure of an example processing template 33, showing in particular the form definition data set 400, in accordance with the invention.
FIG. 5 illustrates a detailed architectural diagram of the intelligent forms processor 2S.
FIG. 6 is a flow diagram of a sequence of operational steps for creating the form definition data set 400.
FIGS. 7, 7a, 7b is a flow diagram of a sequence of operational steps for performing forms recognition process 24A and the field extraction process 24B.
FIG. 8 is an example field, illustrating how it is divided into inner and surrounding regions, in accordance with the invention.
FIG. 9 shows the solution architecture.
FIG. 10 shows strange viewpoints.
FIG. 11 shows the perfect case.
FIG. 12 shows the background text case.
FIG. 13 shows the case of a line in the field's existing inner viewpoint.
FIG. 14 shows the case for lines forming boxes in the field's existing inner viewpoint.
FIG. 15 shows the case for non-straight lines crossing the field.
FIG. 16 shows the case for non-straight lines crossing the field in another example.
FIG. 17 shows the long lines case.
FIG. 18 shows the case of lines or boxes of a form inside the field boundary.
FIG. 19 shows the background text.
FIG. 20 shows background symbols.
FIG. 21 shows the case of an oblique line crossing the inner field mark.
FIG. 22 shows the filled in data which goes beyond the inner field mark.
FIG. 23 shows the case of accepted characters and the separated radicals in a character.
FIG. 24 shows the viewpoint structure.
FIG. 25 shows the viewpoint relations.
FIG. 26A shows the case where every viewpoint connects at most two other viewpoints.
FIG. 26B shows a non-linear example.
FIG. 27 shows examples of lines, corners, crosses and blobs.
FIG. 28 shows erasing.
FIG. 29 shows an example of inside verification given a rectangle, a skew and an offset.
FIG. 30 shows outer field mark boundary generation.
FIG. 31 shows an example of non-straight line tracing.
FIG. 32 shows an example of a split line.
FIG. 33 shows an example coordinate system.
FIG. 34 shows an example of a growing area for a line.
The invention described herein is a data processing system and method for field extraction of scanned images of document forms. The process finds application in a forms processing system described in the copending patent applications by T. S. Betts, et al., cited above. In order to provide some background for the invention disclosed herein, the following description will cover some of the salient features of the T. S. Betts, et al. system. This is further illustrated with reference to FIGS. 1-1A, 2A-2D, 3 and 5.
FIG. 1 shows an overall architectural description of the data processing system for repairing character recognition errors for digital images of document forms. FIG. 1 is partitioned into a left half FIG. 1B and a right half FIG. 1C. A hard copy document 10 is scanned into the system with a scanner 22, resulting in a document image 10' which is shown in FIG. 2C. In the preferred embodiment of the invention, an intelligent forms processor 25, shown in greater detail in the architectural diagram of FIG. 5, provides the form processing template definition processor 57, which generates the processing template 33. The document form processing template 33 specifies the identity and preferred sequence for selected, customized character recognition processes and selected, customized coded data error correction processes. It is these processes which the user feels are reasonably likely to be needed to automatically process a selected batch of documents whose fields have certain, anticipated, uniform characteristics. In accordance with the invention disclosed herein, the processing template 33 shown in FIG. 4 includes the form definition data set 400. It is a form definition data set which is created by the process of the invention disclosed herein and which is utilized by the field extraction process portion of the invention disclosed herein. An example structure for the document form processing template 33 is shown in FIG. 4.
In the preferred embodiment of the invention, the intelligent forms processor 25 of FIG. 5, also provides the scanner interface process with the scanner 22. The intelligent forms processor 25, shown in dotted outline in FIG. 1, also includes a buffer 40, a forms recognition and field extraction processor 24, a character recognition processor 26 and an artificial intelligence error correction processor 28. In the preferred embodiment of the invention, these various processors are implemented through corresponding computer programs which, when executed, perform the respective processes. In an alternate embodiment of the invention, as is shown in FIG. 1, the form processing template definition processor 57, the scanner processor 23, the buffer 40, the forms recognition and field extraction processor 24, the character recognition processor 26 and the artificial intelligence error correction processor 28, may each be separate processors connected together by the local area network (LAN) 20. In order to facilitate an understanding of the invention, the separate processor embodiment shown in FIG. 1 will be described first.
The local area network (LAN) 20 interconnects the template processor 57, scanner processor 23, buffer 40, forms recognition and field extraction processor 24, character recognition processor 26 and artificial intelligence error correction processor 28, to the host computer 34. Also connected to the LAN 20 is the data base error correction processor 30 and the manual verify and correction processor 32. In addition, an additional image capture facility buffer server 40' can be connected to the LAN 20. The host computer 34 has connected to it a document image storage 36, a coded data storage 35, and a repair history storage 38.
When a selected batch of document forms is to be processed by the system, the method of the invention begins by inputting a document form processing template 33 which has been selected by the user. The form definition data set 400 is extracted from the processing template 33, and the field extraction information 404 stored therein is used to perform the field extraction operation, in accordance with the invention. The template will include a first sequence specification for a first plurality of character recognition processes that will be run by the character recognition processor 26. These will include a first occurring and a second occurring character recognition processes. The processing template 33 will also include a second sequence specification for a second plurality of coded data repair processes, such as will be carried out in the artificial intelligence error correction processor 28, or the data base error correction processor 30, or the manual verify and correction processor 32. These processors can include a first occurring and a second occurring coded data repair process.
A document 10 from the batch, is scanned into the system by means of the scanner 22, forming the document image 10' shown in greater detail in FIG. 2C. The document image 10' is a bit map representation of the image of the hard copy document 10. The scanner 22 outputs the document image 10' over the local area network 20 to the forms recognition and field extraction processor 24. A copy of the document image 10' can also be transferred to the host computer 34 and a temporary indexing identity can be assigned to the document image 10' so that it can be temporarily archived in the document image storage 36.
The forms recognition and field extraction processor 24 will include a library of master form definition data sets 400 of forms which are to be processed by the system. The model of a form consists of a form pattern and a description of each field contained on a form. The form pattern is a set of characteristics that are used to distinguish one form type from another. The field description consists of the location of the field on the form expressed in Cartesian coordinates, an acceptance threshold for character recognition, identifying and field specifying information. A data field location can be expressed in the model by two points that describe the opposing corners of a rectangle. The rectangular area may be called a mask. Inspection of FIG. 4 will show that the form definition data set 400 includes form recognition information 402 which is described in greater detail in the copending U.S. patent application by R. G. Casey, et al., cited above. Also, in accordance with the invention, the field extraction information 404 is included in the form data set 400. The field extraction information 404 will include a first field 405 with its field coordinates 406 and its viewpoint structure 408. A second field 405' can be included in the field extraction information 404, which includes its field coordinates 406' and its viewpoint structure 408'. All of the fields on a form which are to be subjected to character recognition, have a field entry 405 and corresponding coordinate entries 406 and viewpoint structure entries 408, in accordance with the invention, in the form definition data set 400.
When the document image 10' is input to the forms recognition and field extraction processor 24, the image is analyzed to identify the form so that the appropriate master form definition data can be accessed. Several methods can be used for form recognition. One method matches the form number which may be typically printed on the document. A second method compares the layout or geography of the forms text and lines which may differentiate form types. A third method relies on the horizontal and vertical lines of the form. Once the document image 10' for the form is characterized, the corresponding master form definition data can be accessed which defines the coordinates for the respective fields, the names of the fields, and the code page for each respective field in the form.
When the scanner 22 scans in the document 10, the scanner processor 23 can perform some front-end operations such as rescaling and compression of the image into an image file to which is assigned a file address. The image file of the document image 10' can be temporarily buffered in the buffer 40 or alternately in the buffer server 40', accessible under an assigned address. Alternately, the scanner processor 23 can output the document image 10' directly to the forms recognition and field extraction processor 24.
The forms recognition and field extraction process 700 shown in FIG. 7 is broken up into the forms recognition process 24A and the field extraction process 24B. This process will be described in greater detail below.
FIG. 2B shows additional information which is included in the master form definition for the master form 10F shown in FIG. 2A. FIG. 2B depicts a master machine generated data structure (MGDS) 50M. The fixed form data 65 is included in the master MGDS 50M, such as the form name "claim," the number of fields and other like information. The identity of related fields, expressed in their crier of priority, is inserted at 77 in the fixed form data portion of the MGDS. In addition, optional space can be provided insert the form processing template 33 in the space 33S. Also included in the fixed form data 65 of the master MGDS 50M of FIG. 2B is the page information PI for each page of the form. In the example shown herein, a single page form is employed. Fixed form data for a page would include for example the page number. Also included in the master MGDS 50M is fixed field data 64 for each of the six data fields in the master form 10F of FIG. 2A. Each field on the form is represented by field data segment, such as field data segment 52M in FIG. 2B which corresponds to the first data field 14F in FIG. 2A. The field data segment 52M includes fixed field data 67 such as the field coordinates 64. Shown in FIG. 2B are the coordinates X1 and Y1 for the upper left-hand corner and X2 and Y2 or the lower right-hand corner of each rectangular field of the master form 10F of FIG. 2A. Also included in the fixed field data 67 is the field name, for example for the field data segment 52M corresponding to the last name field 14F of FIG. 2A, specifies that the field name 68 is "last name." Also included in the fixed field data 67 of FIG. 2B is the field type 7C. For example, the field data segment 52M of FIG. 2B has a field type "A/N" which corresponds to alphanumeric.
Other field data 67 can be included. The master MGDS 50M shown in FIG. 2B is one example of how the master form definition data can be organized and stored in the forms recognition and field extraction processor 24. The forms recognition and field extraction processor 24 will organize the fixed form data 65 for the form information FI and the page information PI and it will organize the fixed field data 67 for each respective field data segment 52M, etc. into a master MGDS 50M which it will transfer to the character recognition processor 26. Prior to such transfer, additional data will be added which is developed by the forms recognition process steps 100 through 102B, and in particular the information which will be added is the skew and offset data 62 for the form image 10', and optionally, the template 22 can be inserted into the space 33S.
Additional spaces can be allocated in the master MGDS 50M of FIG. 2B, for information which will be added at later stages in the recognition process. For example, the form return FR field 72 can have space allocated, optional document image file pointer 60 and optional extracted image file pointer 74 can be allocated. The coded data portion 75 can be added, among others. Each of these allocated fields in the MGDS 50M will be empty until they reach the processing stage in the system where the corresponding data is generated in the character recognition repair of the form.
Reference can be made to FIG. 3 which illustrates an example field extraction process. FIG. 3 shows the field image 16' for the first name field as it appears on the document image of FIG. 2C. The field image 16' is shown with the surrounding box 17, the characters 18 and the spurious lines 19 and 19a. In the process of field extraction, the image of the box 17 is removed. In addition, in the process of extraneous line removal, the image of the extraneous line 19a is eliminated, because a portion of it lies outside the pre-defined area of the field 16'. However, for this example, the extraneous lines 19 have not been removed from the image 18 of the characters in the field 16', as can be seen in the resulting extracted field 16" of FIG. 3. The extracted field image 16" will appear on the extracted field images 10" of FIG. 2D for the form.
An additional example of spurious lines which remain in the extracted field is shown for the state field 12a' in the form image 10' of FIG. 2C, whose corresponding extracted image 12a' still possesses the spurious lines overlapping the letter L in the state abbreviation for the state of Florida "FL."
Then the field extraction process assembles the fixed form data 65, for example the form name 66, etc., along with the fixed field data 67, for example the field name 68, type 70, field coordinates 64, etc., into the master MGDS 50M shown in FIG. 2B. Then, variable form data is inserted into the master MGDS 50M. For example, the form return code 72, which currently has a value of "zero," the document image pointer 60, which is the pointer address for the file of the document image 10', the extracted image pointer 74, which is the pointer address for the file containing the extracted image 10", the skew and offset value 62, and other data currently available at this stage in the forms recognition and field extraction processor 24.
The field extraction process outputs the MGDS message 50A shown in FIG. 1A, to the character recognition processor 26. If this were the multi-task embodiment of these processes in the intelligent forms processor 25 of FIG. 5, then the transfer of the MGDS message 50A would be to a commonly accessible partition in the memory of the processor 25. Alternately in the multiprocessor LAN embodiment of FIG. 1, the MGDS message 50A is transferred over the LAN 20 to the character recognition processor 26.
The field extraction process outputs the extracted images file 10". The extracted field images file 10" can be placed in a file whose address is referred to by the extracted image pointer 74, and the extracted image 10" can then be stored in the buffer 40 or the buffer server 40'. Alternately, the extracted field images 10" can be transferred directly to the character recognition processor 26.
FIG. 1A shows a more detailed illustration of the MGDS 50A as it is output from the forms recognition and field extraction processor 24. In particular, the MGDS 50A shown in FIG. 1A is depicted for the first field data segment 52A and the second field data segment 54A corresponding to the first two out of the six fields of the form image 10' shown in FIG. 2C.
The MGDS 50A shown in FIG. 1A also includes a portion 59 for the storage of the template ID. For the example shown, the template ID is "flood claims," and this will be inserted into portion 59. Also included in the MGDS 50A of FIG. 1A is a portion 76 for storing the field re-route destination (RR). If a particular combination of errors results in the template pointing to a re-route destination, the identity of that re-route destination will be stored in portion 76. Also shown in FIG. 1A is the portion 63 of MGDS 50A, which will store the identity of the recognition routine for each respective field subjected to character recognition in the character recognition processor 26. Also shown in FIG. 1A is the optional inclusion of the processing template 33 in the MGDS 50A. Also, FIG. 1A shows the location of the related fields data 77 in the MGDS 50A.
Optionally, an image data parameter (not shown) may be included in the MGDS message 50A, provided by the processor 24. After the processor 24 has performed its field extraction operation, the cleaned-up image of the field can have its bit map arranged in serial order and optionally compressed and loaded into MGDS 50A.
It is appropriate at this point to discuss how the preferred embodiment of the invention uses a single intelligent forms processor 25 to execute in multi-task form, the respective processes carried out in the forms recognition and field extraction processor 24, the character recognition processor 26, and the artificial intelligence error correction processor 28.
FIG. 5 shows a detailed architectural diagram of the intelligent forms processor 25. The processor 25 includes a memory 302 which is connected by the bus 322 to the CPU 310, the bulk storage device 312, the display, mouse pointing device and keyboard 314, and the LAN adapter 316. The scanner adapter 318 connects the scanner 22 to the bus 322. The LAN adapter 316 connects the LAN 20 to the bus 322. In addition, an optional FAX and communications network adapter 320 can couple a fax telephone line to the bus 322 and can connect a data communications network to the bus 322. The CPU 310, can be for example an Intel 386 or 486 data processing unit. The bulk storage device 312 can be a hard disk storage such as a 200 megabyte hard disk storage, or it can be a combination of a hard disk magnetic storage and an optical read-write storage.
The memory 302 will include several computer programs, each of which is a sequence of executable instructions to carry out a process. The memory 302 includes the form processing template definition program 57A' which carries out the processes 57A performed by the form processing template definition processor 57. In accordance with the invention, the form definition data 400 is created by program 600 of FIG. 6 and inserted into the template 33, shown in FIG. 4. The memory 302 further includes the template assembly partition 33P which is a space set aside for the assembly for the template 33 through the operation of the template definition program 57A'. Also included in the memory 302 is the form processing template 33, which is represented for example as Table 1 and FIG. 4. Still further, a handprint recognition routine, a fast Latin font recognition routine, and an accurate Latin font recognition routine are also provided in the memory 302 for use with the character recognition program 26A'. Still further, an English first name AI error correction routine, an international first name AI error correction routine, and a state abbreviation AI error correction routine are provided in the memory 302, to be used with the AI correction program 28A'. The memory 302 includes a scanner interface program 23A' which carries out the processes performed by the scanner processor 23, as described above. Also included in the memory 302 is a partition for the buffer 40. Still further, a partition is provided in the memory 302 for the master form format 10F which is shown in FIG. 2A and in FIG. 2B.
The memory 302 provides a partition for the forms recognition program 24A' which carries out the forms recognition process 24A. The partition also includes the field extraction program 24B' which carries out the field extraction process 24B of FIG. 7. The result of the execution of the forms recognition program 24A' and the field extraction program 24B' is the production of the extracted field images 10" and the production of the MGDS 50A, as has been described above.
Another partition is provided in the memory 302 for the character recognition program 26A' which carries out the character recognition process 26A.
A partition is also provided in the memory 302 for the artificial intelligence error correction program 28A' which carries out the artificial intelligence process 28A.
A multi-tasking operating system 304, such as IBM's Operating Systems/2, Extended Edition, can be used in multi-tasking mode to control the multi-task execution of the programs in the memory 302. Alternately, the operating system 304 need not oversee the multi-task execution of the application programs, but instead the application programs can be sequentially executed for processing a particular data field segment in the MGDS.
The provision of a facsimile and network adapter 320 connecting a telephone facsimile line to the processor 25, enables the receipt of facsimile images in lieu of images provided through the scanner adapter 318. Similarly, the connection of the adapter 320 to a data communications network enables the receipt of existing document image files from the network, in lieu of the production of such document image files by the scanner 22.
The MGDS message 50C is output from the AI error correction processor 28 over the local area network 20 to the second repair station which is the data base error correction processor 30. The data base error correction processor 30 contains a data base of customer names used in the particular application running on the host computer 34. The host computer 34, in this example, manages a document image archiving system for insurance forms, for an insurance company. The insurance company will have many insured customers and the list of those insured customers will be contained in the data base error correction processor 30. By comparing suggested character strings in the MGDS 50C which is input to the data base error correction processor 30, a selection can be made of that character string which is the same as the actual name of one of the insured customers in the particular host computer application. In this example, the MGDS 50C is input from the AI error correction processor 28 to the data base error correction processor 30 where there is appended thereto a second repair segment R2. The second repair segment R2 will characterize the second repair performed by the processor 30. In an alternate embodiment, the data base error correction processor 30 can be combined with the artificial intelligence error correction processor 28 and the character recognition processor 26 in the same processor 25 in a multi-task application.
The manual verify and correction processor 32 will accept the MGDS 50D and will append to it a third repair segment R3. Depending upon the application, a required manual verification can be made on the result of the sequential correction for the character string which has been flagged for the particular field.
The coordinates for the character or the entire string as the application may require, are abstracted from the reject character information S originally produced by the character recognition processor 26. These are the coordinates of the location in the image bit map for the field and the reject or suspicious character identified by the character recognition processor 26. The correction processor 32 will use these coordinate values to highlight and/or magnifying the field as it is displayed to the operator. A copy of the document image 10' will be delivered over the LAN 20 to the correction processor 32, for viewing by the operator. The location specified by the coordinates of the rejected or suspicious character are used to highlight and/or magnify that portion of the document image displayed. The operator can then quickly compare the highlighted portion of the document image to the character string 48 "John," which is also displayed in alphanumeric characters on the display screen of the processor 32. If the operator accepts the representation of the string 48 "John," then a validation acceptance is recorded in the segment R3. The validation indication can be entered into the segment R3, for example under the comments portion.
Alternately, if the operator wishes to make a change such as capitalizing all of the characters to change string 48 into string 49 "JOHN," then, the string 48 is transferred from the character data buffer B to the input data repair portion 55 of the segment R3. Then the processor 32 loads the corrected value string 49 "JOHN" into the character data buffer B. Other appropriate fields are filled in the segment R3, such as the field name, character position, and comments.
At a later time, if the host computer 34 is instructed to conduct an audit as to how the character string in buffer B for a particular field data segment in the MGDS 50E was repaired, the history of repairs for the character recognition of that field can be ascertained by accessing the MGDS 50E from the repair history storage 38. Each respective repair segment corresponding to the field being audited, can be examined to determine how the repair was carried out. For example, for the second field segment 54 which represents the first name field 16', an audit of the character recognition repairs would examine the field data segment 54, the repair segment R3, the repair segment R2, and the repair segment R1, in that sequential order. This would enable working backwards from the final best estimate of the field contained in the field data segment 54 of the MGDS 50E, backward in reverse sequence for the repairs that were executed on that field. If such an audit were conducted in order to improve the overall repair process, a clear indication as to how each stage of the repair was conducted would be revealed from the collective repair history represented by the repair segments in the MGDS 50E.
The resulting MGDS message 50E now contains a history of the sequential repairs performed on the contents of the first name field 16. The MGDS message 50E can then be transmitted over the LAN 20 to the buffer storage 40' for temporary storage. Alternately it can be transferred to the repair history storage 38 for long term storage. Alternately, it can be transferred to the host computer 34 where the contents of the coded data buffer B for the field data segment 54, which is the string 49, can be loaded into the coded data storage 35. The contents can be used for example as an index value where the application uses the insured customer name as the index value to store the document image 10'.
The form processing template 33 may be stored in the buffer 40 of the intelligent forms processor 25. Alternately, it may be stored in the buffer server 40' of FIG. 1. When a selected batch of document forms is to be processed, the user will select the processing template 33 which specifies the identity and preferred sequence for the customized character recognition processes and customized coded data error correction processes which the user believes are reasonably likely to be needed to automatically process the incoming batch document forms whose fields have certain uniform characteristics anticipated by the user.
Detailed Description of the Process to Create the Form Definition Data Set 400
As was described above, the processing template 33 shown in FIG. 4 is created in the process 57A. Reference can be made to the copending patent application by T. S. Betts, et al. for a detailed description of how the process 57A creates the header portion 78 and the field processing definition portion 80 of the processing template 33 of FIG. 4. The invention disclosed herein deals with the process for forming the definition data set 400 component of the processing template 33. This is shown in the flow diagram of FIG. 6.
The process for creating the form definition data set 400 begins in FIG. 6 with step 600 which prompts the user to input a form name. The form name is input to the system shown in FIG. 5 and then the process continues in step 602 by defining the form recognition features. This portion of the process is described in greater detail in the copending U.S. patent application by R. G. Casey, et al., cited above. Then in step 604 the form recognition features are buffered.
Then in step 606 a loop is started to define the fields. There may be a plurality of fields and they will be identified by an index I which is initially set equal to one. Then in step 608, the process issues a prompt to the user to mark the ith field. Then in step 610 the field viewpoint structure is defined for the ith field. In step 612 the coordinates of the marked field are gotten. Reference can be made here to the example field shown in FIG. 8.
FIG. 8 shows the field 800 which includes a surrounding box 802 and which also includes enclosed boxes 804 and 806 which are separated by a dash line 808. Also in proximity to the box 802 and immediately above it is a body of text 810 which reads "FIELD:TELEPHONE NUMBER." Step 608 of the flow diagram of FIG. 6 results in the positioning of a marked box 812 shown in FIG. 8. The marked box 812 can be generated by the user manipulating a pointing device such as a mouse in the system of FIG. 5, to begin at an upper left-hand coordinate X1,Y1 and to end at a lower right-hand coordinate X2,Y2. The step 612 of the flow of FIG. 6 gets the coordinates X1,Y1 and X2,Y2. Optionally, the coordinates for the other two vertices of the box 812 can also be gotten at this time, namely X3,Y3 for the upper right-hand vortex and X4,Y4 for the lower left-hand vortex. These two additional vertices are optional.
Then step 614 of the flow diagram of FIG. 6 starts a loop to analyze the eight neighbor regions and the inner region for the field 800 of FIG. 8. The surrounding regions are identified as the upper left UL, upper upper UU, upper right UR, the extreme right RR, the lower right or down right DR, the down down DD, the down left, DL and extreme left are LL. These are correspondingly numbered 1, 2, 3, 4, 5, 6, 7 and 8 for these regions in FIG. 8. The enclosed interior region 9 is the region enclosed by the box 812 marked by the user in step 608.
The loop which begins at step 614 will increment through regions 1-9 of field 800 of FIG. 8. For each region, a determination is made in step 616 as to whether there are any background viewpoint primitives within that region. A background viewpoint primitive is a line end, a box end, crossed lines, V-end, T-ends, or blobs. This analysis is described in greater detail below.
The loop then proceeds to step 618 in FIG. 6, to buffer any viewpoint primitive identified for the region defined by the user. The loop increments through all nine regions and after the ninth region is analyzed, step 618 of FIG. 6 flows to step 620 which links any appropriate viewpoint primitives into a view relation for the ith field. An additional description of this portion of the process will be given below. Then the flow diagram of FIG. 6 flows to step 622 which buffers the field viewpoint structure for the ith field. Then in step 624, the index value I is incremented by one and the loop returns to step 608 for all of the fields to be defined by the user. When the user indicates that he has completed defining fields, then the process flows to step 626 which stores the form recognition features defined in step 602, into the form definition data set 400 as the form recognition information 402. Then the flow diagram of FIG. 6 flows to step 628 which stores the coordinates for each field 405 stores the coordinates 406 and the viewpoint structure 408 in the field extraction information 404 of the form definition data set 400. This is stored for each of the fields defined by the user in the process of FIG. 6.
The resulting form definition data set 400 is shown in FIG. 4, and an example of the form definition data set 400, illustrating in detail the field extraction information 404, is shown in Table 1.
Detailed Description of the Field Extraction Process 24B
The forms recognition and field extraction process 700 shown in FIG. 7 is broken into the forms recognition portion 24A and the field extraction portion 24B.
The forms recognition portion 24A of FIG. 7 starts after step 700 which scans in the filled-out form. Then step 702 performs the forms recognition process. This process is described in greater detail in the copending U.S. patent application by R. G. Casey, et al. cited above.
Then step 703 outputs the identity of the form as the form ID and step 704 outputs the skew and the offset data for the particular scanned-in form. The skew is defined as the error in the Y direction or the scanning direction which is the motion of the document as it went through the scanner, divided by the error in the X direction which is perpendicular to the Y direction. This is represented as delta Y1 divided by delta X1. The offset is the displacement of the scanned-in form from its expected nominal position and the Y offset value in the direction of motion of the form is delta Y2 and the horizontal or perpendicular offset is delta X2. These values of skew and offset are then output to the field extraction process 24B.
The field extraction process 24B of FIG. 7 begins with step 710 and flows to step 712 which gets the field definition information from the forms definition data set, using the form ID output in step 703. Then step 714 starts a loop for incrementing through all of the fields on the form. The index value of N is set equal to one. Then in step 716 the field mark coordinates are extracted from the definition data set. Then in step 718, the adjusted coordinates are computed for the offset of the form. These are the coordinates for each respective field. For example the field 800 shown in FIG. 8 will have the coordinate X1 adjusted by adding to it the value delta X2 and it will have the coordinate Y1 adjusted by adding to it the value delta Y2 from the offset data output in step 704. This is done for X1 and Y1 and for X2 and Y2 in the field 800 of FIG. 8. Optionally, it can also be done for X3,Y3 and X4,Y4.
Then step 720 of FIG. 7 starts the loop to compare the stored viewpoint structure in the form definition data set 400 with the scanned-in viewpoint structure for each respective field, in this case the nth field. Then step 722 starts another loop which cycles through the nine regions 1-9 indicated in FIG. 8, determining if any background viewpoint primitive exists in the ith region of the form definition data set field. This would be a determination as to whether there is a line end, box end, or other viewpoint primitive for the particular region under investigation, the ith region. Then step 724 determines if there is a background viewpoint primitive found, then it is compared with any viewpoint primitive in the corresponding ith region of the scanned-in image of the field currently under investigation. This comparison will then identify the type of viewpoint primitive found in the scanned field image and it will compare the relative location of the viewpoint primitive in the scanned image with the expected location of the viewpoint primitive in the form definition data set. The type of the viewpoint primitive is buffered and the relative location is buffered in step 724. Then step 726 will increment the index value of I for the regions 1-9 until all nine regions indicated in FIG. 8 have been analyzed. Then the process of FIG. 7 flows to step 728 which removes from the scanned image of the field under investigation, those viewpoint primitives which match the stored viewpoint primitives in the form definition data set 400.
Then the flow diagram of FIG. 7 flows to step 730 which performs a partial removal from the scanned image of viewpoint structures called strange viewpoints. This form of structure is described in greater detail in the discussion below.
Then step 732 starts fine tuning the ambiguous pixels in the remaining field image by checking for overlapped pixels. Where there are overlapped pixels, the pixel at the overlapped point is retained. This is described in greater detail below. Then in step 734 a check is made for any long line segments and they are removed. This step is further described below. Then in step 736, a check is made for jagged, connected lines and they are removed. This step is further described below. Then in step 738, a check is made for isolated pixels and use is made of the skew data output in step 704, and the distance from the adjusted field mark, to determine the validity of the isolated pixel. If the isolated pixel is determined to be valid, then it is retained. This step is described further below.
Then step 740 extracts the remainder of the field image of the scanned-in form for the field under investigation, and buffers it in the system of FIG. 5. Step 742 increments the field index of N to the next value and loops back to step 716, until all of the fields in the form have been processed. Then the flow diagram flows to step 744 which outputs the extracted field to the character recognition process. Optionally, step 746 can also output the coordinates of the adjusted field mark which may be used by the character recognition process and which also can be stored as the modified coordinates for the field in the MGDS, as is shown in FIG. 1A, where the field coordinate 64 is adjusted in its value to become the adjusted field coordinates output by step 746.
Table 1 gives a specific example of the field extraction information 404 generated by the field extraction process 24B of FIG. 7, when it is applied to the field 800 shown in FIG. 8. The field mark coordinates 406 are shown as X1,Y1 and X1,Y2 in Table 1. The remaining portion of Table 1 represents the viewpoint structure 408 for the field 800 shown in FIG. 8. The viewpoint structure is divided into the nine regions 1-9 shown in FIG. 8 and sequentially set forth in Table 1. Region 1 shows a viewpoint primitive type as box end 1 and that corresponds to the box end 820 of FIG. 8. The coordinates for the corner of the box end 820 are X11,Y11 and are also included in Table 1.
Region 2 is set forth in Table 1 with two viewpoint primitives, the first one is a blob type and the second one is a line type. For the first viewpoint primitive, the text 810 is characterized as a blob type viewpoint primitive and a circumscribing box is defined for a blob type primitive as described below. The coordinates of the circumscribing box are included in the region 2 blob viewpoint primitive definition. The second viewpoint primitive for region 2 is a line type and its coordinates are also given. The line viewpoint primitive is 822 in FIG. 8.
Region 3 has a box end 2 type viewpoint primitive 824. Region 4 has a line type primitive 826. Region 5 has a box end primitive 828. Region 6 has a line type primitive 830. Region 7 has a box type primitive 832. Region 8 has a line type primitive 834. Region 9 which is inside the marked box 812, has nine viewpoint primitives, four viewpoint primitives are box end types for the box 804. One viewpoint primitives is a line type primitive for the line 808. Four additional viewpoint primitives are box end type primitives for the box 806. These all correspond to region 9.
Table 1 is constructed by the process carried in the flow diagram of FIG. 6 which identifies and constructs for each defined field by the user, the corresponding viewpoint structure. If the process to create the form definition data set of FIG. 6 is applied to the field 800 shown in FIG. 8, the result would be the field extraction information 404 represented by Table 1 with the coordinates 406 for the marked area and the viewpoint structure 408.
When a form which includes the field 800 is scanned into the system for the purpose of performing character recognition on the filled-in information within the box 800, the forms recognition and field extraction process of FIG. 7 will get the form definition data set 400 represented by Table 1 and will perform the field extraction operation using the definitions of Table 1 to extract the portion of the field within the box 812 in FIG. 8. ##SPC1##
TABLE 1__________________________________________________________________________406 FIELD MARK COORDINATES X1 Y1 X2 Y2 OPTIONAL406 FIELD VIEWPOINT STRUCTURE X3 Y3 COORD. CORNER X4 Y4 TYPEREGION 1 - BOX END1 X11, Y11REGION 2 - BLOB CIRCUMSCRIBING BOX COORD. (X21; Y21) (X22; Y22) BIT MAP IMAGE OF BLOB LINE (X23; Y23) (224; Y24) COORD. CORNERREGION 3 - BOX END2 (X31; Y31)REGION 4 - LINE (X41; Y41) (X42; Y42)REGION 5 - BOX END3 (X41; Y51)REGION 6 - LINE (X61; Y61) (X62; Y62)REGION 7 - BOX END4 (X71; Y71)REGION 8 - LINE (X81; Y81) (X82; Y82)INSIDE REGION 9 - BOX END1 (X9,1; Y9,1) 2 9,2 3 9,3 4 9,4 LINE X95,1; Y13,1) (X96-Y13,2) BOX END1 14 2 15 3 16 4 (X1,1; Y1,1)__________________________________________________________________________
Although a specific embodiment of the invention has been disclosed, it will be understood by those having skill in the art that changes can be made to that specific embodiment without departing from the spirit and the scope of the invention.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4020462 *||Dec 8, 1975||Apr 26, 1977||International Business Machines Corporation||Method and apparatus for form removal from contour compressed image data|
|US4741045 *||Sep 23, 1983||Apr 26, 1988||Dest Corporation||Optical character isolation system, apparatus and method|
|US4813077 *||Jul 30, 1986||Mar 14, 1989||Scan-Optics, Inc.||Sales transaction record processing system and method|
|US4949392 *||May 20, 1988||Aug 14, 1990||Eastman Kodak Company||Document recognition and automatic indexing for optical character recognition|
|US5010580 *||Aug 25, 1989||Apr 23, 1991||Hewlett-Packard Company||Method and apparatus for extracting information from forms|
|US5025484 *||Aug 15, 1988||Jun 18, 1991||Kabushiki Kaisha Toshiba||Character reader device|
|US5050222 *||May 21, 1990||Sep 17, 1991||Eastman Kodak Company||Polygon-based technique for the automatic classification of text and graphics components from digitized paper-based forms|
|US5093873 *||Dec 27, 1990||Mar 3, 1992||Kabushiki Kaisha Toshiba||Electronic filing apparatus using part of read image information as retrieval information|
|US5123062 *||Aug 5, 1991||Jun 16, 1992||Kabushiki Kaisha Toshiba||OCR for sequentially displaying document layout according to recognition process|
|US5140650 *||Feb 2, 1989||Aug 18, 1992||International Business Machines Corporation||Computer-implemented method for automatic extraction of data from printed forms|
|US5199081 *||Dec 14, 1990||Mar 30, 1993||Kabushiki Kaisha Toshiba||System for recording an image having a facial image and id information|
|US5229589 *||Nov 21, 1991||Jul 20, 1993||Optimum Solutions Corp., Inc.||Questionnaire scanning system employing expandable answer mark areas for efficient scanning and mark detection|
|CA2000012A1 *||Oct 2, 1989||Aug 2, 1990||Richard G. Casey||Computer-implemented method for automatic extraction of data from printed forms|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US5692073 *||Jul 30, 1996||Nov 25, 1997||Xerox Corporation||Formless forms and paper web using a reference-based mark extraction technique|
|US5748809 *||Jun 23, 1997||May 5, 1998||Xerox Corporation||Active area identification on a machine readable form using form landmarks|
|US5751842 *||Nov 15, 1996||May 12, 1998||Ncr Corporation||Document transaction apparatus|
|US5768416 *||Jun 7, 1995||Jun 16, 1998||Millennium L.P.||Information processing methodology|
|US5815595 *||Dec 29, 1995||Sep 29, 1998||Seiko Epson Corporation||Method and apparatus for identifying text fields and checkboxes in digitized images|
|US5822454 *||Apr 10, 1995||Oct 13, 1998||Rebus Technology, Inc.||System and method for automatic page registration and automatic zone detection during forms processing|
|US5872640 *||Nov 22, 1996||Feb 16, 1999||Cable & Wireless, Inc.||Facsimile form generation system|
|US6009194 *||Jul 7, 1997||Dec 28, 1999||International Business Machines Corporation||Methods, systems and computer program products for analyzing information in forms using cell adjacency relationships|
|US6009195 *||Jun 12, 1996||Dec 28, 1999||Minolta Co., Ltd.||Image processor with marker edition mode|
|US6044375 *||Apr 30, 1998||Mar 28, 2000||Hewlett-Packard Company||Automatic extraction of metadata using a neural network|
|US6094505 *||Mar 19, 1998||Jul 25, 2000||Millennium L.P.||Information processing methodology|
|US6154754 *||Sep 25, 1997||Nov 28, 2000||Siemens Corporate Research, Inc.||Automatic synthesis of semantic information from multimedia documents|
|US6201894 *||Jan 22, 1997||Mar 13, 2001||Canon Kabushiki Kaisha||Method and apparatus for extracting ruled lines or region surrounding ruled lines|
|US6338093||Jun 12, 2000||Jan 8, 2002||Dirienzo Andrew L.||Attachment integrated claims system and operating method therefor|
|US6343310||Dec 26, 2000||Jan 29, 2002||Dirienzo Andrew L.||Attachment integrated claims system and operating method therefor|
|US6400845||Apr 23, 1999||Jun 4, 2002||Computer Services, Inc.||System and method for data extraction from digital images|
|US6683697||Dec 9, 1999||Jan 27, 2004||Millenium L.P.||Information processing methodology|
|US6741737||Mar 18, 1999||May 25, 2004||Choicepoint Asset Company||System and method for the secure data entry from document images|
|US6943923 *||Dec 4, 2000||Sep 13, 2005||Itesoft S.A.||Adaptive technology for automatic document analysis|
|US6970267 *||Jan 8, 2002||Nov 29, 2005||Scan-Optics Inc.||Gray scale optical mark reader|
|US7075673||Nov 6, 2003||Jul 11, 2006||Eon-Net L.P.||Information processing methodology|
|US7142728||May 17, 2002||Nov 28, 2006||Science Applications International Corporation||Method and system for extracting information from a document|
|US7149347||Mar 2, 2000||Dec 12, 2006||Science Applications International Corporation||Machine learning of document templates for data extraction|
|US7178020||Jul 17, 2002||Feb 13, 2007||Integrated Claims Systems, Llc||Attachment integrated claims system and operating method therefor|
|US7184162||Apr 15, 2005||Feb 27, 2007||Eon-Net L.P.||Information processing methodology|
|US7212301 *||Oct 31, 2001||May 1, 2007||Call-Tell Llc||System and method for centralized, automatic extraction of data from remotely transmitted forms|
|US7251380 *||Apr 1, 2003||Jul 31, 2007||Abbyy Software Ltd.||Adjustment method of a machine-readable form model and a filled form scanned image thereof in the presence of distortion|
|US7259887||Apr 10, 2006||Aug 21, 2007||Eon-Net L.P.||Information processing methodology|
|US7346768||May 25, 2006||Mar 18, 2008||Integrated Claims Systems, Llc||Attachment integrated claims systems and operating methods therefor|
|US7409632||Sep 25, 2001||Aug 5, 2008||Integrated Claims Systems, Llc||Classifying, disabling and transmitting form fields in response to data entry|
|US7424672 *||Oct 3, 2003||Sep 9, 2008||Hewlett-Packard Development Company, L.P.||System and method of specifying image document layout definition|
|US7426486||Oct 31, 2001||Sep 16, 2008||Call-Tell Llc||Multi-party reporting system and method|
|US7474434||Mar 16, 2007||Jan 6, 2009||Millennium L.P.||Information processing methodology|
|US7508537 *||May 1, 2007||Mar 24, 2009||Call-Tell Llc||Centralized, automatic extraction of data from remotely transformed and transmitted forms|
|US7561734||Oct 23, 2006||Jul 14, 2009||Science Applications International Corporation||Machine learning of document templates for data extraction|
|US7570383||Mar 16, 2007||Aug 4, 2009||Glory Licensing Llc||Information processing methodology|
|US7570811 *||Nov 20, 2003||Aug 4, 2009||Oce Technologies B.V.||Segmenting an image via a graph|
|US7619768||Oct 31, 2007||Nov 17, 2009||Glory Licensing Llc||Information processing methodology|
|US7672007||Oct 31, 2007||Mar 2, 2010||Glory Licensing Llc||Information processing methodology|
|US7694129||Dec 29, 2006||Apr 6, 2010||Integrated Claims Systems, Llc||Methods of processing digital data and images|
|US7764830||Oct 23, 2006||Jul 27, 2010||Science Applications International Corporation||Machine learning of document templates for data extraction|
|US7881561 *||Jun 26, 2003||Feb 1, 2011||Abbyy Software Ltd.||Method of pre-analysis of a machine-readable form image|
|US7916946 *||Jul 10, 2007||Mar 29, 2011||Canon Kabushiki Kaisha||Image processing apparatus for identifying and classifying an object in an image, method of controlling same, and computer-readable storage medium storing a computer program for controlling same|
|US7930292 *||Jan 24, 2006||Apr 19, 2011||Canon Kabushiki Kaisha||Information processing apparatus and control method thereof|
|US8155979||Jan 5, 2007||Apr 10, 2012||Integrated Claim Systems, LLC||Attachment integrated claims systems and operating methods therefor|
|US8170371 *||Jun 26, 2003||May 1, 2012||Abbyy Software, Ltd||Method of image pre-analyzing of a machine-readable form of non-fixed layout|
|US8171391||Nov 3, 2006||May 1, 2012||Abbyy Software, Ltd||Method of describing the structure of graphical objects|
|US8571262||Sep 8, 2010||Oct 29, 2013||Abbyy Development Llc||Methods of object search and recognition|
|US8660294||Sep 27, 2006||Feb 25, 2014||Charactell Ltd.||Form data extraction without customization|
|US8676609||Jan 12, 2007||Mar 18, 2014||Integrated Claims Systens, LLC||Attachment integrated claims systems and operating methods therefor|
|US8750571||Aug 9, 2013||Jun 10, 2014||Abbyy Development Llc||Methods of object search and recognition|
|US8805093 *||Dec 22, 2010||Aug 12, 2014||Abbyy Development Llc||Method of pre-analysis of a machine-readable form image|
|US8908969||Jul 31, 2012||Dec 9, 2014||Abbyy Development Llc||Creating flexible structure descriptions|
|US9015573||Apr 17, 2012||Apr 21, 2015||Abbyy Development Llc||Object recognition and describing structure of graphical objects|
|US9111140 *||Jan 10, 2012||Aug 18, 2015||Dst Technologies, Inc.||Identification and separation of form and feature elements from handwritten and other user supplied elements|
|US9224040||Sep 23, 2011||Dec 29, 2015||Abbyy Development Llc||Method for object recognition and describing structure of graphical objects|
|US9262679||Jul 22, 2015||Feb 16, 2016||Dst Technologies, Inc.||System and method for identification and separation of form and feature elements from handwritten and other user supplied elements|
|US9477567 *||Mar 19, 2014||Oct 25, 2016||Testplant, Inc.||Method for monitoring a graphical user interface on a second computer display from a first computer|
|US9619841||Nov 25, 2014||Apr 11, 2017||Integrated Claims Systems, Llc||Systems to assist in the creation, transmission, and processing of health insurance claims|
|US9633257||Jun 25, 2014||Apr 25, 2017||Abbyy Development Llc||Method and system of pre-analysis and automated classification of documents|
|US9659213 *||Oct 16, 2015||May 23, 2017||Cognizant Technology Solutions India Pvt. Ltd.||System and method for efficient recognition of handwritten characters in documents|
|US9740692||Nov 5, 2014||Aug 22, 2017||Abbyy Development Llc||Creating flexible structure descriptions of documents with repetitive non-regular structures|
|US20020194035 *||Jul 17, 2002||Dec 19, 2002||Dirienzo Andrew L.||Attachment integrated claims system and operating method therefor|
|US20030083966 *||Oct 31, 2001||May 1, 2003||Varda Treibach-Heck||Multi-party reporting system and method|
|US20030084402 *||Oct 31, 2001||May 1, 2003||Varda Treibach-Heck||Remote, automatic reporting system and method|
|US20030108243 *||Dec 4, 2000||Jun 12, 2003||Didier Charpentier||Adaptive technology for automatic document analysis|
|US20030215137 *||May 17, 2002||Nov 20, 2003||Janusz Wnek||Method and system for extracting information from a document|
|US20040070793 *||Nov 6, 2003||Apr 15, 2004||Millennium, L.P.||Information processing methodology|
|US20040140992 *||Nov 20, 2003||Jul 22, 2004||Marquering Henricus A.||Segmenting an image via a graph|
|US20040163042 *||Feb 17, 2004||Aug 19, 2004||Itm Associates||Method and system for annotating documents using an independent annotation repository|
|US20040190790 *||Jun 26, 2003||Sep 30, 2004||Konstantin Zuev||Method of image pre-analyzing of a machine-readable form of non-fixed layout|
|US20050067482 *||Sep 26, 2003||Mar 31, 2005||Wu Daniel Huong-Yu||System and method for data capture and management|
|US20050076295 *||Oct 3, 2003||Apr 7, 2005||Simske Steven J.||System and method of specifying image document layout definition|
|US20050089248 *||Apr 1, 2003||Apr 28, 2005||Konstantin Zuev||Adjustment method of a machine-readable form model and a filled form scanned image thereof in the presence of distortion|
|US20050185218 *||Apr 15, 2005||Aug 25, 2005||Eon-Net L.P.||Information processing methodology|
|US20060173904 *||Jan 24, 2006||Aug 3, 2006||Canon Kabushiki Kaisha||Information Processing Apparatus and Control Method Thereof|
|US20060181742 *||Apr 10, 2006||Aug 17, 2006||Millenium L.P.||Information processing methodology|
|US20060274941 *||Jun 26, 2003||Dec 7, 2006||Konstantin Zuev||Method of pre-analysis of a machine-readable form image|
|US20070011449 *||May 25, 2006||Jan 11, 2007||Dirienzo Andrew L||Attachment integrated claims systems and operating methods therefor|
|US20070053611 *||Oct 10, 2006||Mar 8, 2007||Janusz Wnek||Method and system for extracting information from a document|
|US20070188802 *||Mar 16, 2007||Aug 16, 2007||Millenium L.P.||Information processing methodology|
|US20070195370 *||Sep 22, 2006||Aug 23, 2007||Satoshi Suga||Workflow system, server device, and processing method of workflow system|
|US20070198596 *||Jan 5, 2007||Aug 23, 2007||Dirienzo Andrew L||Attachment integrated claims systems and operating methods therefor|
|US20070203837 *||Dec 29, 2006||Aug 30, 2007||Dirienzo Andrew L||Exchange and integration of mixed data formats|
|US20070203914 *||Dec 29, 2006||Aug 30, 2007||Dirienzo Andrew L||Attachment integrated claims systems and operating methods therefor|
|US20070206223 *||May 1, 2007||Sep 6, 2007||Varda Treibach-Heck||Centralized, automatic extraction of data from remotely transformed and transmitted forms|
|US20070260493 *||Jan 12, 2007||Nov 8, 2007||Dirienzo Andrew L||Attachment integrated claims systems and operating methods therefor|
|US20080025606 *||Jul 10, 2007||Jan 31, 2008||Canon Kabushiki Kaisha||Image processing apparatus, method of controlling same, and program|
|US20080109403 *||Nov 3, 2006||May 8, 2008||Konstantin Zuev||Method of describing the structure of graphical objects.|
|US20080155456 *||Mar 10, 2008||Jun 26, 2008||Integrated Claims Systems, Llc;||Attachment integrated claims systems and operating methods therefor|
|US20080180736 *||Oct 31, 2007||Jul 31, 2008||Eon-Net L.P.||Information processing methodology|
|US20080285792 *||Sep 27, 2006||Nov 20, 2008||Ofer Comay||Form Data Extraction Without Customization|
|US20080292175 *||Oct 31, 2007||Nov 27, 2008||Eon-Net L.P.||Information processing methodology|
|US20090132477 *||Nov 3, 2006||May 21, 2009||Konstantin Zuev||Methods of object search and recognition.|
|US20110013806 *||Sep 8, 2010||Jan 20, 2011||Abbyy Software Ltd||Methods of object search and recognition|
|US20110091109 *||Dec 22, 2010||Apr 21, 2011||Abbyy Software Ltd||Method of pre-analysis of a machine-readable form image|
|US20110167331 *||Feb 28, 2011||Jul 7, 2011||Altman Ian K||Method and system for annotating documents using an independent annotation repository|
|US20110188759 *||Apr 14, 2011||Aug 4, 2011||Irina Filimonova||Method and System of Pre-Analysis and Automated Classification of Documents|
|US20130177246 *||Jan 10, 2012||Jul 11, 2013||Kevin W. Stokes||Identification and Separation of Form and Feature Elements from Handwritten and Other User Supplied Elements|
|US20140208170 *||Mar 19, 2014||Jul 24, 2014||Testplant, Inc.||Method for monitoring a graphical user interface on a second computer display from a first computer|
|US20150039707 *||Mar 19, 2013||Feb 5, 2015||Intuit Inc.||Document processing|
|US20150261735 *||Mar 11, 2015||Sep 17, 2015||Ricoh Company, Ltd.||Document processing system, document processing apparatus, and document processing method|
|US20160307067 *||Jun 29, 2016||Oct 20, 2016||Abbyy Development Llc||Method and apparatus for determining a document type of a digital document|
|EP1946233A2 *||Sep 27, 2006||Jul 23, 2008||Charactell Ltd.||Form data extraction without customization|
|EP1946233A4 *||Sep 27, 2006||Feb 27, 2013||Charactell Ltd||Form data extraction without customization|
|WO1997020427A1 *||Nov 26, 1996||Jun 5, 1997||Cable & Wireless, Inc.||Facsimile form generation system|
|U.S. Classification||382/173, 382/317, 358/403, 715/210, 358/453|
|International Classification||G06K9/20, G06K9/00|
|Cooperative Classification||G06K2209/01, G06K9/2054|
|Oct 21, 1992||AS||Assignment|
Owner name: IBM CORPORATION, NEW YORK
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNOR:HUANG, SHIOW L.;REEL/FRAME:006370/0209
Effective date: 19921021
|Sep 4, 1998||FPAY||Fee payment|
Year of fee payment: 4
|Sep 19, 2002||FPAY||Fee payment|
Year of fee payment: 8
|Nov 29, 2006||REMI||Maintenance fee reminder mailed|
|May 16, 2007||LAPS||Lapse for failure to pay maintenance fees|